Machine learning-based damage classification and comparative life cycle assessment of Origami Pill Bug for emergency shelters

Angshuman C. Baruah, Ann C. Sychterz,
Machine learning-based damage classification and comparative life cycle assessment of Origami Pill Bug for emergency shelters,
Journal of Building Engineering,
Volume 113,
2025,
114051,
ISSN 2352-7102,
https://doi.org/10.1016/j.jobe.2025.114051.
(https://www.sciencedirect.com/science/article/pii/S2352710225022880)
Abstract: There is a critical need for transformative emergency shelter solutions that integrate rapid deployment capabilities with structural resilience and sustainability in response to increasing severe weather events. While deployable origami structures offer geometric transformability, compact transportation, and reusability advantages, implementation remains limited due to absence of frameworks unifying structural reliability with quantitative environmental evaluation. This research establishes a reproducible framework integrating experimental structural performance analysis, machine learning-based damage detection, and comparative life cycle assessment for sustainable emergency shelter design. A meter-scale Origami Pill Bug prototype was experimentally characterized across healthy and damaged scenarios. Statistical analysis with Mann–Whitney U tests and Holm–Bonferroni correction established significant strain variations enabling damage detection baselines. Three supervised machine learning algorithms were evaluated, with Support Vector Machines achieving optimal performance through superior accuracy and convergence. Comparative life cycle assessment quantified environmental impacts against conventional alternatives using ISO 14044 standards. Results demonstrate that statistical analysis of strain variations due to damage enables machine learning-based damage detection capabilities essential for multi-use scenarios, while revealing substantial environmental benefits with the Origami Pill Bug shelter achieving 37.8% Global Warming Potential reduction for single-use and 68.7% reduction over five cycles compared to typical emergency shelters. This integrated framework advances deployable origami structures as viable sustainable emergency infrastructure solutions by providing a novel comprehensive methodology linking damage detection-enabled reusability with quantified environmental performance, establishing foundations for next-generation disaster response systems.
Keywords: Deployable structures; Damage detection; Sustainability assessment; Machine learning; Sustainable design